OpenRig explained: coordinate coding agents without losing the operator
OpenRig quickstart: preview before `rig up`
Prepare one selected provider and see what startup will change
What you will learn
- Check the actual prerequisites
- Use a plan, then a small rig
- Send one bounded request
Before you start
- Node 22 or 24 and tmux on a supported host
- One working provider account and a disposable repository
Turn the first owner/checker run into evidence for a safe rollout decision
Key takeaways
- A dry run exposes intended machine changes.
- A running daemon is not the same as a ready seat.
- A sent message is not a recorded queue task.
Check the actual prerequisites
This pinned revision requires Node.js 22 or 24 and tmux on macOS or Linux. Native Windows is unsupported and WSL2 untested; Apple Silicon users are directed to Node 22. Verify only the provider CLI and account you intend to use.
The README recommends `rig setup --dry-run` before a broader machine setup. `rig setup` itself may attempt both providers and cmux, whereas the selected-provider path can avoid tools you do not need.
Use a plan, then a small rig
The first-use guide previews the selected starter with `rig specs preview` and then `rig up ... --plan`. A plan is a read-only preview of intended actions; running `rig up` afterward launches the seats and the kernel.
After launch, `rig ps --nodes --rig <name>` checks whether the project seats are actually ready. Resolve authentication and trust prompts before sending work. An accepted CLI command is not evidence that an agent has started.
Send one bounded request
The README sends a concrete task to `dev-owner@first-project` and asks that owner to track the work in the queue and request a check. `rig send` alone does not create a queue item.
Keep the first task local and reversible, then inspect the resulting files and the checker report. No model run, tmux session or setup command was executed for this article.
Decision guide
| Criterion | Option A | Option B |
|---|---|---|
| Best when | You need predictable behavior and easy auditing | You need adaptive optimization and have reliable telemetry |
| Main risk | May leave performance on the table | Can become difficult to explain or debug |
Implementation steps
- 1
Verify Node, tmux and the chosen provider account.
- 2
Preview setup, RigSpec and `rig up` before applying.
- 3
Check both seats, then send one bounded task.
Copy-ready example
node --version
tmux -V
rig setup --dry-run
rig specs preview first-project --kind rig
rig up first-project --cwd . --planFrequently asked questions
Can I evaluate on native Windows?
The pinned README says native Windows is unsupported and WSL2 untested.
Does `rig send` enqueue work?
No. The recipient must record the task in the queue.
Sources
- OpenRig / README.mdSource checked 2026-10-04
- OpenRig / docs/reference/getting-started.mdSource checked 2026-10-04
- OpenRig / packages/cli/src/commands/setup.tsSource checked 2026-10-04
- OpenRig / packages/cli/src/commands/up.tsSource checked 2026-10-04